Guiding Attention in Sequence-to-Sequence Models for Dialogue Act Prediction
| Autor: | Matteo Manica, Giovanna Varni, Emmanuel Vignon, Pierre Colombo, Emile Chapuis, Chloé Clavel |
|---|---|
| Přispěvatelé: | IBM France Lab [Biot], IBM - Paris [Bois-Colombes], IBM-IBM, Laboratoire Traitement et Communication de l'Information (LTCI), Institut Mines-Télécom [Paris] (IMT)-Télécom Paris, Télécom Paris |
| Jazyk: | angličtina |
| Rok vydání: | 2020 |
| Předmět: |
Conditional random field
FOS: Computer and information sciences Computer Science - Machine Learning Machine translation Computer science media_common.quotation_subject Inference 02 engineering and technology computer.software_genre Machine learning Machine Learning (cs.LG) 0202 electrical engineering electronic engineering information engineering Leverage (statistics) Conversation [INFO]Computer Science [cs] Dialog box ComputingMilieux_MISCELLANEOUS media_common Computer Science - Computation and Language business.industry General Medicine 020201 artificial intelligence & image processing Artificial intelligence business computer Computation and Language (cs.CL) |
| Zdroj: | Proceedings of the AAAI Conference on Artificial Intelligence Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34 (05), pp.7594-7601. ⟨10.1609/aaai.v34i05.6259⟩ AAAI |
| ISSN: | 2159-5399 |
| DOI: | 10.1609/aaai.v34i05.6259⟩ |
| Popis: | The task of predicting dialog acts (DA) based on conversational dialog is a key component in the development of conversational agents. Accurately predicting DAs requires a precise modeling of both the conversation and the global tag dependencies. We leverage seq2seq approaches widely adopted in Neural Machine Translation (NMT) to improve the modelling of tag sequentiality. Seq2seq models are known to learn complex global dependencies while currently proposed approaches using linear conditional random fields (CRF) only model local tag dependencies. In this work, we introduce a seq2seq model tailored for DA classification using: a hierarchical encoder, a novel guided attention mechanism and beam search applied to both training and inference. Compared to the state of the art our model does not require handcrafted features and is trained end-to-end. Furthermore, the proposed approach achieves an unmatched accuracy score of 85% on SwDA, and state-of-the-art accuracy score of 91.6% on MRDA. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|